Tom Engsted

TY - UNPB

T1 - Statistical vs. Economic Significance in Economics and Econometrics: Further comments on McCloskey & Ziliak

AU - Engsted, Tom

PY - 2009

Y1 - 2009

N2 - I comment on the controversy between McCloskey & Ziliak andHoover & Siegler on statistical versus economic significance, inthe March 2008 issue of the Journal of Economic Methodology.I argue that while McCloskey & Ziliak are right in emphasizing'real error', i.e. non-sampling error that cannot be eliminatedthrough specification testing, they fail to acknowledge those areasin economics, e.g. rational expectations macroeconomics andasset pricing, where researchers clearly distinguish between statisticaland economic significance and where statistical testingplays a relatively minor role in model evaluation. In these areasmodels are treated as inherently misspecified and, consequently,are evaluated empirically by other methods than statistical tests.I also criticise McCloskey & Ziliak for their strong focus on thesize of parameter estimates while neglecting the important questionof how to obtain reliable estimates, and I argue that significancetests are useful tools in those cases where a statisticalmodel serves as input in the quantification of an economic model.Finally, I provide a specific example from economics - asset returnpredictability - where the distinction between statistical andeconomic significance is well appreciated, but which also showshow statistical tests have contributed to our substantive economicunderstanding.

AB - I comment on the controversy between McCloskey & Ziliak andHoover & Siegler on statistical versus economic significance, inthe March 2008 issue of the Journal of Economic Methodology.I argue that while McCloskey & Ziliak are right in emphasizing'real error', i.e. non-sampling error that cannot be eliminatedthrough specification testing, they fail to acknowledge those areasin economics, e.g. rational expectations macroeconomics andasset pricing, where researchers clearly distinguish between statisticaland economic significance and where statistical testingplays a relatively minor role in model evaluation. In these areasmodels are treated as inherently misspecified and, consequently,are evaluated empirically by other methods than statistical tests.I also criticise McCloskey & Ziliak for their strong focus on thesize of parameter estimates while neglecting the important questionof how to obtain reliable estimates, and I argue that significancetests are useful tools in those cases where a statisticalmodel serves as input in the quantification of an economic model.Finally, I provide a specific example from economics - asset returnpredictability - where the distinction between statistical andeconomic significance is well appreciated, but which also showshow statistical tests have contributed to our substantive economicunderstanding.

KW - Statistical and economic significance, statistical hypothesis testing, model evaluation, misspecified models

M3 - Working paper

BT - Statistical vs. Economic Significance in Economics and Econometrics: Further comments on McCloskey & Ziliak

PB - Institut for Økonomi, Aarhus Universitet

CY - Aarhus

ER -